The Consequences of Divorce 1275 But does divorce lower people's well-being,or are following divorce.That is,long-standing differ- poorly functioning people especially likely to di- ences between those who divorce and those who vorce?Consistent with the divorce-stress-adjust- remain married might be amplified as divorce be- ment perspective,and contrary to the selection comes imminent.For example,a husband's ag- perspective,longitudinal studies show that people gression might contribute to the dissolution of the who make the transition from marriage to divorce relationship,but the dissolution of the relation- report an increase in symptoms of depression,an ship,in turn,might generate even more serious increase in alcohol use,and decreases in happi- levels of aggression.Second,some individuals ness,mastery,and self-acceptance (Aseltine might have long-standing problems that disrupt Kessler,1993;Hope,Rodgers,Power,1999; their marriages,whereas others might be relatively Marks Lambert,1998;Power,Rodgers, symptom-free until confronted with the stress of Hope,1999).Given that divorce is a process rath- marital dissolution.In other words,selection ex- er than a discrete event,declines in well-being are planations might apply to some groups of people likely to begin prior to the legal divorce.In fact, more than others. Kitson's (1992)respondents reported (retrospec- tively)that they had experienced the greatest level Divorce as Crisis or Chronic Strain? of stress prior to making the decision to divorce the second highest level of distress at the time of An unresolved issue in the literature of the 1990s the decision,and the least stress following the fi- is whether divorce represents a temporary crisis nal separation.Consistent with Kitson's data,lon- to which most individuals adapt or a source of gitudinal studies (Booth Amato.1991:Johnson chronic strains that persist indefinitely.Several Wu.1996:Mastekaasa.1994b.1997)show that studies found that unhappiness,distress,depres- reports of unhappiness and psychological distress sion,alcohol consumption,and health problems begin to rise a few years prior to marital separa- had largely subsided 2 or 3 years aftersepara- tion.Furthermore.Johnson and Wu (1996)used a tion-a result that supports the crisis model fixed-effects model to control for all time-invari- (Booth Amato,1991;Goldberg,Greenberger, ant individual variables,thus making it unlikely Hamill.O'Neil.1992:Kitson.1992:Lorenz et that selection could account for the increase in al..1997).In contrast,other studies failed to find distress. improvements in people's functioning during the Some longitudinal studies,however,suggest time since divorce.unless they remarried-a re- that selection effects operate alongside divorce ef- sult that supports the chronic strain model (Asel- fects.For example,Mastekaasa (1997)observed tine Kessler,1993:DeGarmo Kitson.1996: personal problems(such as greater alcohol con- Gray Silver,1990;Johnson Wu,1996;Mas- sumption among wives)as early as 4 years prior tekaasa,1995;Neff Schluter,1993;Wang to divorce.Hope,Rodgers,and Power (1999) Amato,in press).Furthermore,Umberson and found that depression at age 23 predicted becom- Williams (1993)found that parental strain among ing a single mother at age 33.Similarly,Davies divorced fathers increased,rather than decreased. and colleagues (1997)found that many divorced over time.Of course,both the crisis and the mothers had a history of depression that predated chronic strain models might contain some truth the marriage.These mothers also reported high Kitson(1992)found that although half of her re- levels of adversity in their families of origin,in- spondents improved over time,about one fourth cluding weak attachment to parents and parental got worse.These results suggest that a crisis mod- depression.Controlling for these family-of-origin el(implying gradual adjustment)might be appro- factors decreased the estimated effect of divorce priate for some individuals,and a chronic strain on adult depression(suggesting a selection effect), model (implying persistent long-term problems) although the association between divorce and de- might be appropriate for others. pression remained significant(suggesting divorce causation). In general,studies support the notion of di- Mediators of Divorce Effects vorce causation,but a degree of selection also Researchers attempting to identify the mediators might be operating.This combination can occur of divorce effects have adopted two strategies. in two ways.First,some individuals might be Some researchers have examined associations be- prone to psychological or interpersonal problems tween mediators and measures of well-being using prior to divorce but exhibit additional problems samples composed entirely of divorced individu-The Consequences of Divorce 1275 But does divorce lower people’s well-being, or are poorly functioning people especially likely to di￾vorce? Consistent with the divorce-stress-adjust￾ment perspective, and contrary to the selection perspective, longitudinal studies show that people who make the transition from marriage to divorce report an increase in symptoms of depression, an increase in alcohol use, and decreases in happi￾ness, mastery, and self-acceptance (Aseltine & Kessler, 1993; Hope, Rodgers, & Power, 1999; Marks & Lambert, 1998; Power, Rodgers, & Hope, 1999). Given that divorce is a process rath￾er than a discrete event, declines in well-being are likely to begin prior to the legal divorce. In fact, Kitson’s (1992) respondents reported (retrospec￾tively) that they had experienced the greatest level of stress prior to making the decision to divorce, the second highest level of distress at the time of the decision, and the least stress following the fi- nal separation. Consistent with Kitson’s data, lon￾gitudinal studies (Booth & Amato, 1991; Johnson & Wu, 1996; Mastekaasa, 1994b, 1997) show that reports of unhappiness and psychological distress begin to rise a few years prior to marital separa￾tion. Furthermore, Johnson and Wu (1996) used a fixed-effects model to control for all time-invari￾ant individual variables, thus making it unlikely that selection could account for the increase in distress. Some longitudinal studies, however, suggest that selection effects operate alongside divorce ef￾fects. For example, Mastekaasa (1997) observed personal problems (such as greater alcohol con￾sumption among wives) as early as 4 years prior to divorce. Hope, Rodgers, and Power (1999) found that depression at age 23 predicted becom￾ing a single mother at age 33. Similarly, Davies and colleagues (1997) found that many divorced mothers had a history of depression that predated the marriage. These mothers also reported high levels of adversity in their families of origin, in￾cluding weak attachment to parents and parental depression. Controlling for these family-of-origin factors decreased the estimated effect of divorce on adult depression (suggesting a selection effect), although the association between divorce and de￾pression remained significant (suggesting divorce causation). In general, studies support the notion of di￾vorce causation, but a degree of selection also might be operating. This combination can occur in two ways. First, some individuals might be prone to psychological or interpersonal problems prior to divorce but exhibit additional problems following divorce. That is, long-standing differ￾ences between those who divorce and those who remain married might be amplified as divorce be￾comes imminent. For example, a husband’s ag￾gression might contribute to the dissolution of the relationship, but the dissolution of the relation￾ship, in turn, might generate even more serious levels of aggression. Second, some individuals might have long-standing problems that disrupt their marriages, whereas others might be relatively symptom-free until confronted with the stress of marital dissolution. In other words, selection ex￾planations might apply to some groups of people more than others. Divorce as Crisis or Chronic Strain? An unresolved issue in the literature of the 1990s is whether divorce represents a temporary crisis to which most individuals adapt or a source of chronic strains that persist indefinitely. Several studies found that unhappiness, distress, depres￾sion, alcohol consumption, and health problems had largely subsided 2 or 3 years aftersepara￾tion—a result that supports the crisis model (Booth & Amato, 1991; Goldberg, Greenberger, Hamill, & O’Neil, 1992; Kitson, 1992; Lorenz et al., 1997). In contrast, other studies failed to find improvements in people’s functioning during the time since divorce, unless they remarried—a re￾sult that supports the chronic strain model (Asel￾tine & Kessler, 1993; DeGarmo & Kitson, 1996; Gray & Silver, 1990; Johnson & Wu, 1996; Mas￾tekaasa, 1995; Neff & Schluter, 1993; Wang & Amato, in press). Furthermore, Umberson and Williams (1993) found that parental strain among divorced fathers increased, rather than decreased, over time. Of course, both the crisis and the chronic strain models might contain some truth. Kitson (1992) found that although half of her re￾spondents improved over time, about one fourth got worse. These results suggest that a crisis mod￾el (implying gradual adjustment) might be appro￾priate for some individuals, and a chronic strain model (implying persistent long-term problems) might be appropriate for others. Mediators of Divorce Effects Researchers attempting to identify the mediators of divorce effects have adopted two strategies. Some researchers have examined associations be￾tween mediators and measures of well-being using samples composed entirely of divorced individu-